This was a hybrid event with in-person attendance in Wu and Chen and virtual attendance…
Autonomous systems of the near future have the potential to impact and improve our lives in unprecedented ways. As an example, an autonomous taxi service can be instructed to take the socially responsible route minimizing congestion (and associated carbon emissions) instead of the purely selfish shortest route. In order for such a system to be robust and gain public acceptance, it is critical that it respects the users’ personal incentives, as well as ensures that different users enjoy equitable service. Until now, the vast majority of incentive schemes used in engineering systems have been monetary. Examples include road-tolling to reduce traffic congestion, surge-pricing to manage excessive demand in ride-hailing, and time-varying energy tariffs to encourage consumption off peak hours. In this talk, we argue that the use of money in engineering systems is fundamentally flawed. The sensitivity of users to money can vary widely for factors outside the system (e.g., income and inter-generational wealth accumulation), and therefore monetary schemes implicitly favor the wealthy. We introduce the concept of a self-contained karma economy as an alternative incentive scheme. In a karma economy, each user is endowed with tokens (called karma) that are non-tradable for money, and the users repeatedly obtain or yield a preferential service in exchange of karma. We demonstrate this concept on the stylized yet insightful example of autonomous intersection management, where an agent passing first in an intersection transfers karma to that yielding. We lay the mathematical foundations that ensure that such an economy is well-defined and the strategic behavior of its users (i.e., how much karma they spend) can be predicted. We then analyze the performance of the karma economy in numerical case studies, showcasing that it achieves socially efficient outcomes in a self-contained, and thereby equitable, manner.